Overview There are several measurements for right ventricular (RV) function in 2D echocardiography, for example Fractional Area Change (FAC), Tricuspid Annular Plane Systolic Excursion (TAPSE) and Right Ventricular Index of Myocardial Performance (RIMP). The limitations of 2D measurements for RV assessment are well-known. Foreshortening and geometric assumptions make 2D not well suited for quantification. 3D Auto RV — right ventricular quantification Anke Sedlmeier, Senior Product Manager, TOMTEC David Prater, MS, Clinical Scientist, Philips Ultrasound White paper The importance of RV function has been underestimated due to enormous attention given to the evaluation of the left heart, lack of familiarity with ultrasound techniques that can be used in imaging the right heart, and the complexity of the RV shape and anatomy (Figure 1). These limitations can be overcome by using 3D. With the growing availability of 3D techniques such as magnetic resonance (MR) imaging and 3D echocardiography, assessment of shape and functional changes in the right ventricle becomes feasible. Three-dimensional modalities offer exceptional accuracy and reproducibility in the assessment of RV shape and function. Figure 1 Compared to the left ventricle (left), the right ventricle shape (right) is much more complex.
8
Embed
3D Auto RV — right ventricular quantification · 4 Muraru D, et al. New speckle-tracking algorithm for right ventricular volume analysis from three-dimensional echocardiographic
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
OverviewThere are several measurements for right ventricular (RV) function in 2D echocardiography, for example Fractional Area Change (FAC), Tricuspid Annular Plane Systolic Excursion (TAPSE) and Right Ventricular Index of Myocardial Performance (RIMP). The limitations of 2D measurements for RV assessment are well-known. Foreshortening and geometric assumptions make 2D not well suited for quantification.
3D Auto RV —right ventricular quantification
Anke Sedlmeier, Senior Product Manager, TOMTEC
David Prater, MS, Clinical Scientist, Philips
Ultrasound
White paper
The importance of RV function has been underestimated
due to enormous attention given to the evaluation of the
left heart, lack of familiarity with ultrasound techniques
that can be used in imaging the right heart, and the
complexity of the RV shape and anatomy (Figure 1).
These limitations can be overcome by using 3D.
With the growing availability of 3D techniques
such as magnetic resonance (MR) imaging and 3D
echocardiography, assessment of shape and functional
changes in the right ventricle becomes feasible.
Three-dimensional modalities offer exceptional
accuracy and reproducibility in the assessment
of RV shape and function.
Figure 1 Compared to the left ventricle (left), the right ventricle shape (right) is much more complex.
2
RV auto-segmentation
Auto-segmentation technology is based on a deformable
model which consists of a boundary detector and multiple
shape models of the RV. The boundary detector uses
information from both image gradients and the distribution
of the image grayscale values. The shape models consist
of a mean shape and shape modes, with the shape modes
accounting for the shape variations due to differences in
anatomy and pathology. The auto-segmentation process
begins with an initial boundary estimate used to select
the best-fitting shape model. The final boundary is a
balance between the initial boundary and the selected
shape model. Both the boundary detection and the
shape models utilize artificial intelligence and are trained
using a large number of clinical images annotated by
echocardiographic experts.
3D Auto RV maintains all the manual functionality of the
TOMTEC 4D RV-FUNCTION software, including View
Adjustment and Tracking Revision. Auto-segmentation
does these two steps automatically but the user may still
do them manually. This functionality provides the speed
and robustness of automatic operation with the flexibility
of manual adjustment.
Although MR has excellent intra- and inter-observer
variability, it is expensive, not widely available, and cannot
be performed in patients with special conditions such as
metallic implanted devices, claustrophobia and irregular
heart rhythm.
Three-dimensional echocardiography is available and well
validated against the gold standard MR. The latest guideline
recommends 3D echocardiography measurements of RV
when knowledge of RV volumes may be clinically important.
TOMTEC has extensive experience in 3D RV measurements.
4D RV-FUNCTION has been part of the TOMTEC portfolio
since 2006. Evolving from 4D RV FUNCTION, the 3D Auto
RV enhanced workflow and performance is powered by
machine learning artificial intelligence, making it feasible
for routine clinical use. This on-cart availability and its
automated workflow within the application accelerate
the process of measuring RV values.
3D echo quantification on right ventricle
The 3D Auto RV application combines proven TOMTEC
experience in RV analysis software with Philips machine
learning artificial intelligence similar to HeartModel A.I..
Auto-segmentation provides the initial image alignment
and contour positioning. These operations significantly
reduce, and in many cases eliminate, the manual image
adjustments needed for RV analysis. The result is both
faster analysis and improved reproducibility.
3
3D Auto RV operation
The 3D Auto RV application is initiated via a touchpanel
button labelled 3D Auto RV. The auto-segmentation process
is executed and the user is shown the Tracking Revision
workflow step (Figure 2). If the user is satisfied with the
tracking, they can proceed directly to the Analysis workflow
step where the dynamic 3D volume (Beutel) is shown along
with the volume waveform and numerical results. For easy
navigation, the work-step list is shown on the side of the
View Adjustment and Tracking Revision workflow steps.
A minimum of 10 volumes per second are required to ensure
the quality of volume analysis.
The entry point for the standard TTE acquisition in
automatic mode is the Tracking Revision workflow step.
This screen shows the RV model and corresponding
2D contours over the complete heart cycle.
Figure 2 Tracking revision workflow step.
The model is tracked over the entire cardiac cycle using
speckle tracking technology. The resulting dynamic
surface model can optionally be adjusted by the user in
end-diastole (ED) or end-systolic (ES). To edit, the user just
drags the contour line in ED or ES. There are different pen
sizes depending on the area the user wants to change.
If more editing is needed, it is highly recommended to start
editing in ED and re-track. If initial model fails completely,
landmarks in the view adjustment state should be checked
and adjusted.
When leaving the application out of Tracking Revision and
activating the “Accept Measurements” checkbox, all global
values are saved.
54
View adjustment
In the View Adjustment workflow step, the user can
manually adjust the image landmarks (Figure 3).
At end-diastole, the left and right ventricular long axes
are set to define a basic coordinate system. Based on that,
a RV-focused four-chamber view and a short-axis view
are derived by the system. Also, the heart cycle to be
analyzed can be changed in this state.
In the View Adjustment workflow step, the following
general settings can be adjusted:
• Default acquisition window (TTE, subcostal)
• Automatic start in tracking revision
(only available for TTE)
• Change default heart cycle
• Change default pause time (pause time in view
adjustment to set landmarks)
• View orientation (apex up/down)
• Change default layout (in tracking revision)
• Result display size
Figure 3 View Adjustment.
Analysis
3D volumes over time are computed numerically
from the dynamic surface model and used to calculate
What does the ‘x’ marker in the view adjustment indicate?
The cross in the SAX view shows where the Apex-MV and
Apex-TV axes intersect the SAX. This view can be helpful
in orienting the views.
Frequently asked questions
How does the software define ED and ES time points;
can they be edited?
ED and ES are defined as time points of largest (ED)
and smallest (ES) volume. The time points are defined
after tracking in the Tracking Revision workflow step.
The blue flags of the electrocardiogram (ECG) are
positioned accordingly.
They cannot be edited.
How does 3D Auto RV detect the borders? Does it use
speckle tracking?
The boundary detection utilizes information from both
image gradient and the distribution of the image grayscale
values. The shape models consist of a mean shape and
shape modes, with the shape modes accounting for shape
variations due to differences in anatomy and pathology.
The model is tracked over the entire cardiac cycle using
speckle tracking technology.
What landmarks does it use for tracking?
No individual landmarks are used for tracking, but rather
sections of the 3D model contour.
What is 3D Auto RV validated against?
Refer to validation section.
Are we including trabeculae or excluding them?
Trabeculae are included in the model volume of RV.
How does it work on dilated hearts?
Dilated hearts are validated in publications. The challenge
with dilated hearts is to acquire images such that the free
wall is in the field of view.
How does 3D Auto RV work on contrast images?
Due to the small amount of available data sets, there
is no validation for contrast data available.
Are there any issues with lower EF’s? Hypercontractility?
There are no known issues.
Figure 6 SAX view with pulmonary valve side-cutting.
76
References
1 Leary PJ, et al. Three-dimensional analysis of right ventricular shape and function in pulmonary hypertension. Pulm Circ. 2012 Jan-Mar;2(1):34–40.
doi: 10.4103/2045-8932.94828.
2 Lang RM, Badano LP, et al. Recommendations for Cardiac Chamber Quantification by Echocardiography in Adults: An Update from the American Society
of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr. 2015 Jan;28(1):1-39. doi: 10.1016/j.echo.2014.10.003 .
3 Messner AM, Taylor GQ. Algorithm 550, Solid Polyhedron Measures. ACM Transactions on Mathematical Software. 1980 Mar;6(1):121-130.
4 Muraru D, et al. New speckle-tracking algorithm for right ventricular volume analysis from three-dimensional echocardiographic data sets: validation
with cardiac magnetic resonance and comparison with the previous analysis tool. Eur Heart J Cardiovasc Imaging. 2016 Nov;17(11):1279-1289.
5 Laser KT, Karabiyik A, Korperich H, Horst JP, Barth P, Kececioglu D, Burchert W, Dalla Pozza R, Herberg, U. Validation and Reference Values for
Three-Dimensional Echocardiographic Right Ventricular Volumetry in Children: A Multicenter Study. J Am Soc Echocardiogr. 2018 Sep;31(9):1050-1063.
doi: 10.1016/j.echo.2018.03 .010.
6 Medvedofsky D, et al. Novel Approach to Three-Dimensional Echocardiographic Quantification of Right Ventricular Volumes and Function
from Focused Views. J Am Soc Echocardiogr. 2015 Oct;28(10):1222-31. doi: 10.1016/j.echo.2015.06. 013.
What’s the recommendation if it doesn’t track?
If the tracking fails but the orientation of the data set is
correct, then in the tracking revision stage the contour in
the ED frame should be edited and the “re-track” button
used. ED will be re-tracked.
If the complete model orientation failed, e.g., Apex and
MV are interchanged, go to the View Adjustment workflow
step. Correct the landmark position of the Apex (LV), MV,
Apex (RV) and TV and proceed with tracking revision.
Will it reject a volume? If so, what are the parameters
in which it will reject?
A minimum of 10 volumes per heart cycle and 10 volumes
per second is required.
If the results of 3D Auto RV vary when scanning
the same patient, how can I explain that?
The algorithm is searching for contours in the acquired
image. If the quality differs, the contours are different,
e.g., the free wall can be seen completely on one data
set but parts are missing on the second data set.
The algorithm needs to follow its trained knowledge
where the contour of the free wall might be – this can
result in different volume values.
Can I do a multi-beat acquisition and which beat
will 3D Auto RV use?
3D Auto RV is always analyzing one heartbeat at a time.
The heartbeat can be selected in View Adjustment
workflow step. It is possible to pre-select any heartbeat